Zero handling for CLR-transformation of relative abundance from MetaPhlAn

Hi,

To apply relative abundance data from MetaPhlAn into a model, I’d like to use a CLR-transformation or natural log transformation. I would be very happy if I could know any recommendations for zero-handling before the transformation (e.g., min-half replacement or shift, +1 shift to the estimated number of reads).

Best,
Ryohei